building tools, building community
1/29/2015
The rOpenSci project
Packages
Contributors
Community
- ROpenSci Blog: ropensci.org/blog
- Community Calls: github.com/ropensci/commcalls
- UnConference: unconf.ropensci.org
- ROpenSci Github: github.com/ropensci
- ROpenSci Twitter: @ rOpenSci
Goals
- Increase the availability and quality of R packages to interface with research data
- Support reproducibility along the whole data analysis pipeline
- Sustainable software design
- Sustainable community
Remote sensors
credit: NASA
micro sensors
credit: NSF
NEON
credit: Hopkin (2006) doi: 10.1038/444420a
OOI
credit: Witze (2013) doi: 10.1038/501480a
Computer simulations
credit: NERSC
Field-based study
credit: Scambos & Bauer, NSIDC
Growth of climate data by type
Overpeck+ (2011) doi: 10.1126/science.1197869
The real challenge is doing science today that is relevant to the data tomorrow
Example: Synthetic analysis of biodiversity loss
Synthesizes over 140 data sets.
Finds no evidence for systematic loss
How easy would it be to update this to reflect new data?
What is "a lot" of data?
Engineering bottlenecks
Baraniuk (2011) doi: 10.1126/science.1197448
From bottlenecks to workflows
adapted from Reichman+ (2011) doi: 10.1126/science.1197962
From bottlenecks to workflows
Reichman+ (2011) doi: 10.1126/science.1197962
Example workflows: Dynamic documents
Peng (2011) doi: 10.1126/science.1213847